I am working on the following problem:

Show that if $A$ is free on a set $S$ in Ab, with map $\mathrm{\Phi}:S \rightarrow A$, then $A$ is also a coproduct of $|S|$ copies of $\mathbb{Z}$, where each $\mathrm{\theta}_x :\mathbb{Z} \rightarrow A$ is given by $\mathrm{\theta}_x(n)=n\cdot \mathrm{\Phi}(x)$.

I can't use the fact that $A$ is free on $S$ and the definition of a coproduct to prove that $A$ is isomorphic with the coproduct of $|S|$ copies of $\mathrm{Z}$. How do you construct the diagram? I only have to work with the definitions of free object and coproduct.


$X$ is a set. $F$ with $\phi :X \to F$ is free on $X$ in the category $\mathcal{C}$ iff $\forall A \in \mathcal{O}(\mathcal{C})$ and $\psi :X\to A$, $\exists! f \in Mor(F,A)$ such that $f\circ \phi = \psi$.

Suppose $\{A_i : i\in \mathcal{I}\}$ is an indexed family in $\mathcal{O}(\mathcal{C})$. A coproduct of the $A_i$ is an object $A$ together with morphisms $\iota_i \in Mor(A_i,A)$ for all $i\in \mathcal{I}$ satisfying the property: if $B\in \mathcal{O}(\mathcal{C})$, and $\phi_i \in Mor(A_i,B)$ for all $i\in \mathcal{I}$ then $\exists !\tau:A \to B$ such that all the following diagrams are commutative: $$ \begin{array}[c]{ccc} A_i&\xrightarrow{\iota_i}&A\\ &\llap{\phi_i}\searrow& \big\downarrow\rlap{\tau}\\ &&B \end{array}$$

  • 2
    $\begingroup$ It's pretty strange that you must prove something without hypotheses. Every proof coming to my mind would use at the end the universal property of $\coprod$; anyways, you can argue this way: $A\cong F(S)$, where $F\colon {\bf Set}\to {\bf Ab}$ is the free abelian group functor, left adjoint ot the forgetful $U\colon {\bf Ab}\to {\bf Set}$; now since $$ {\bf Ab}(\coprod_{s\in S}\mathbb{Z},B)\cong \prod_{s\in S}B $$ naturally in $B$, since RHS is precisely the set of functions $S\to B$, and by uniqueness of adjoints, you get $F(S)\cong \mathbb{Z}^{(S)}$ $\endgroup$ – Fosco Loregian Jun 3 '16 at 7:03
  • $\begingroup$ Thanks for the answer. I am not very familiar with category theory because I am studying only the basics in my homological algebra class. I have a solution and I would appreciate it if you would check it. $\endgroup$ – user53970 Jun 13 '16 at 17:13
  • $\begingroup$ I changed the problem a little to include the definitions I can use. $\endgroup$ – user53970 Jun 13 '16 at 18:12

$\begin{array}[c]{ccc} A&\stackrel{\Phi}{\longleftarrow}&S\\ \Big\uparrow\rlap{\phi_x}&\searrow &\Big\downarrow\rlap{\xi}\\ \mathbb{Z}_x&\stackrel{\iota_x}{\longrightarrow}&\bigoplus\mathbb{Z}_x \end{array}$

Lets define $\phi_x \in Mor(\mathbb{Z}_x , A)$, so that $\phi_x(n)=n\cdot\Phi(x)$. Since $\bigoplus\mathbb{Z}_x$ is a coproduct, by definition we have $\Rightarrow \exists !f\in Mor(\bigoplus\mathbb{Z}_x,A)$ such that $f\iota_x=\phi_x$.

Now define $\xi(x)=\iota_x(1)$. Since $A$ is free on $S$, by definition we have $\Rightarrow \exists !g\in Mor(A,\bigoplus\mathbb{Z}_x)$ such that $g\Phi = \xi$.

Lets prove that $A\cong \bigoplus\mathbb{Z}_x$. For this we need that $fg=gf=id_A$:

For all $a\in A$, $\exists x\in S$ such that $\Phi(x)=a$ because $\Phi$ is one-to-one. Hence: $$fg(a)=fg\Phi(x)=f\xi(x)=f\iota_x(1)=\phi_x(1)=\Phi(x)=a$$ We concluded that $fg=id_A$. Now, for all $z\in \bigoplus\mathbb{Z}_x$, $\exists x\in S$ and $\exists n\in\mathbb{Z}_x$ such that $z=\iota_x(n)$. So: $$gf(z)=gf\iota_x(n)=g\phi_x(n)=n\cdot g\Phi(x)=n\cdot\xi(x)=n\cdot \iota_x(1)=\iota_x(n)=z$$ So $gf=id_A$ and with this we are done.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.